An airborne short-pulse radar system to measure ice thickness was designed and operated during the 1973–74 and 1974–75 ice seasons. This system supported a joint effort among NASA, the US Coast Guard, and NOAA to develop an all-weather Great Lakes Ice Information System which aids in extending the winter navigation season. This paper describes experimental studies into the accuracy and limitations of this short-pulse radar system. A low-power version of this system was built and operated from an all-terrain vehicle on the Straits of Mackinac during March 1975. The vehicle allowed rapid surveying of large areas and eliminated the ambiguity in location between the radar system and the “ground truth” ice auger team which is unavoidable for the airborne versions of the system. These in situ measurements also permitted an assessment of the effects of snow cover, surface melt water, pressure ridging, and ice type upon the accuracy of the system. Over 25 sites were explored which had ice thicknesses in the range from 29 to 60 cm. The maximum radar overestimate was 9.8%, while the maximum underestimate was 6.6%. The average error of the 25 measurements was 0.1%.